Variability in commercially available deformable image registration: A multi-institution analysis using virtual head and neck phantoms.

PURPOSE: The purpose of this study was to evaluate the performance of three common deformable image registration (DIR) packages across algorithms and institutions. METHODS AND MATERIALS: The Deformable Image Registration Evaluation Project (DIREP) provides ten virtual phantoms derived from computed tomography (CT) datasets of head-and-neck cancer patients over a single treatment course. Using the DIREP phantoms, DIR results from 35 institutions were submitted using either Velocity, MIM, or Eclipse. Submitted deformation vector fields (DVFs) were compared to ground-truth DVFs to calculate target registration error (TRE) for six regions of interest (ROIs). Statistical analysis was performed to determine the variability between each DIR software package and the variability of users within each algorithm. RESULTS: Overall mean TRE was 2.04 ± 0.35 mm for Velocity, 1.10 ± 0.29 mm for MIM, and 2.35 ± 0.15 mm for Eclipse. The MIM mean TRE was significantly different than both Velocity and Eclipse for all ROIs. Velocity and Eclipse mean TREs were not significantly different except for when evaluating the registration of the cord or mandible. Significant differences between institutions were found for the MIM and Velocity platforms. However, these differences could be explained by variations in Velocity DIR parameters and MIM software versions. CONCLUSIONS: Average TRE was shown to be <3 mm for all three software platforms. However, maximum errors could be larger than 2 cm indicating that care should be exercised when using DIR. While MIM performed statistically better than the other packages, all evaluated algorithms had an average TRE better than the largest voxel dimension. For the phantoms studied here, significant differences between algorithm users were minimal suggesting that the algorithm used may have more impact on DIR accuracy than the particular registration technique employed. A significant difference in TRE was discovered between MIM versions showing that DIR QA should be performed after software upgrades as recommended by TG-132.

Commissioning an in-room mobile CT for adaptive proton therapy with a compact proton system.

PURPOSE: To describe the commissioning of AIRO mobile CT system (AIRO) for adaptive proton therapy on a compact double scattering proton therapy system. METHODS: A Gammex phantom was scanned with varying plug patterns, table heights, and mAs on a CT simulator (CT Sim) and on the AIRO. AIRO-specific CT-stopping power ratio (SPR) curves were created with a commonly used stoichiometric method using the Gammex phantom. A RANDO anthropomorphic thorax, pelvis, and head phantom, and a CIRS thorax and head phantom were scanned on the CT Sim and AIRO. Clinically realistic treatment plans and nonclinical plans were generated on the CT Sim images and subsequently copied onto the AIRO CT scans for dose recalculation and comparison for various AIRO SPR curves. Gamma analysis was used to evaluate dosimetric deviation between both plans. RESULTS: AIRO CT values skewed toward solid water when plugs were scanned surrounded by other plugs in phantom. Low-density materials demonstrated largest differences. Dose calculated on AIRO CT scans with stoichiometric-based SPR curves produced over-ranged proton beams when large volumes of low-density material were in the path of the beam. To create equivalent dose distributions on both data sets, the AIRO SPR curve's low-density data points were iteratively adjusted to yield better proton beam range agreement based on isodose lines. Comparison of the stoichiometric-based AIRO SPR curve and the "dose-adjusted" SPR curve showed slight improvement on gamma analysis between the treatment plan and the AIRO plan for single-field plans at the 1%, 1 mm level, but did not affect clinical plans indicating that HU number differences between the CT Sim and AIRO did not affect dose calculations for robust clinical beam arrangements. CONCLUSION: Based on this study, we believe the AIRO can be used offline for adaptive proton therapy on a compact double scattering proton therapy system.

Validating Dose Uncertainty Estimates Produced by AUTODIRECT: An Automated Program to Evaluate Deformable Image Registration Accuracy.

Deformable image registration is a powerful tool for mapping information, such as radiation therapy dose calculations, from one computed tomography image to another. However, deformable image registration is susceptible to mapping errors. Recently, an automated deformable image registration evaluation of confidence tool was proposed to predict voxel-specific deformable image registration dose mapping errors on a patient-by-patient basis. The purpose of this work is to conduct an extensive analysis of automated deformable image registration evaluation of confidence tool to show its effectiveness in estimating dose mapping errors. The proposed format of automated deformable image registration evaluation of confidence tool utilizes 4 simulated patient deformations (3 B-spline-based deformations and 1 rigid transformation) to predict the uncertainty in a deformable image registration algorithm's performance. This workflow is validated for 2 DIR algorithms (B-spline multipass from Velocity and Plastimatch) with 1 physical and 11 virtual phantoms, which have known ground-truth deformations, and with 3 pairs of real patient lung images, which have several hundred identified landmarks. The true dose mapping error distributions closely followed the Student t distributions predicted by automated deformable image registration evaluation of confidence tool for the validation tests: on average, the automated deformable image registration evaluation of confidence tool-produced confidence levels of 50%, 68%, and 95% contained 48.8%, 66.3%, and 93.8% and 50.1%, 67.6%, and 93.8% of the actual errors from Velocity and Plastimatch, respectively. Despite the sparsity of landmark points, the observed error distribution from the 3 lung patient data sets also followed the expected error distribution. The dose error distributions from automated deformable image registration evaluation of confidence tool also demonstrate good resemblance to the true dose error distributions. Automated deformable image registration evaluation of confidence tool was also found to produce accurate confidence intervals for the dose-volume histograms of the deformed dose.

The Mobius AIRO mobile CT for image-guided proton therapy: Characterization & commissioning.

PURPOSE: The purpose of this study was to characterize the Mobius AIRO Mobile CT System for localization and image-guided proton therapy. This is the first known application of the AIRO for proton therapy. METHODS: Five CT images of a Catphan® 504 phantom were acquired on the AIRO Mobile CT System, Varian EDGE radiosurgery system cone beam CT (CBCT), Philips Brilliance Big Bore 16 slice CT simulator, and Siemens SOMATOM Definition AS 20 slice CT simulator. DoseLAB software v.6.6 was utilized for image quality analysis. Modulation transfer function, scaling discrepancy, geometric distortion, spatial resolution, overall uniformity, minimum uniformity, contrast, high CNR, and maximum HU deviation were acquired. Low CNR was acquired manually using the CTP515 module. Localization accuracy and CT Dose Index were measured and compared to reported values on each imaging device. For treatment delivery systems (Edge and Mevion), the localization accuracy of the 3D imaging systems were compared to 2D imaging systems on each system. RESULTS: The AIRO spatial resolution was 0.21 lp mm-1 compared with 0.40 lp mm-1 for the Philips CT Simulator, 0.37 lp mm-1 for the Edge CBCT, and 0.35 lp mm-1 for the Siemens CT Simulator. AIRO/Siemens and AIRO/Philips differences exceeded 100% for scaling discrepancy (191.2% and 145.8%). The AIRO exhibited higher dose (>27 mGy) than the Philips CT Simulator. Localization accuracy (based on the MIMI phantom) was 0.6° and 0.5 mm. Localization accuracy (based on Stereophan) demonstrated maximum AIRO-kV/kV shift differences of 0.1 mm in the x-direction, 0.1 mm in the y-direction, and 0.2 mm in the z-direction. CONCLUSIONS: The localization accuracy of AIRO was determined to be within 0.6° and 0.5 mm despite its slightly lower image quality overall compared to other CT imaging systems at our institution. Based on our study, the Mobile AIRO CT system can be utilized accurately and reliably for image-guided proton therapy.

Benchmarking of five commercial deformable image registration algorithms for head and neck patients.

Benchmarking is a process in which standardized tests are used to assess system performance. The data produced in the process are important for comparative purposes, particularly when considering the implementation and quality assurance of DIR algorithms. In this work, five commercial DIR algorithms (MIM, Velocity, RayStation, Pinnacle, and Eclipse) were benchmarked using a set of 10 virtual phantoms. The phantoms were previously developed based on CT data collected from real head and neck patients. Each phantom includes a start of treatment CT dataset, an end of treatment CT dataset, and the ground-truth deformation vector field (DVF) which links them together. These virtual phantoms were imported into the commercial systems and registered through a deformable process. The resulting DVFs were compared to the ground-truth DVF to determine the target registration error (TRE) at every voxel within the image set. Real treatment plans were also recalculated on each end of treatment CT dataset and the dose transferred according to both the ground-truth and test DVFs. Dosimetric changes were assessed, and TRE was correlated with changes in the DVH of individual structures. In the first part of the study, results show mean TRE on the order of 0.5 mm to 3 mm for all phan-toms and ROIs. In certain instances, however, misregistrations were encountered which produced mean and max errors up to 6.8 mm and 22 mm, respectively. In the second part of the study, dosimetric error was found to be strongly correlated with TRE in the brainstem, but weakly correlated with TRE in the spinal cord. Several interesting cases were assessed which highlight the interplay between the direction and magnitude of TRE and the dose distribution, including the slope of dosimetric gradients and the distance to critical structures. This information can be used to help clinicians better implement and test their algorithms, and also understand the strengths and weaknesses of a dose adaptive approach.

Near Real-Time Assessment of Anatomic and Dosimetric Variations for Head and Neck Radiation Therapy via Graphics Processing Unit-based Dose Deformation Framework.

PURPOSE: The purpose of this study was to systematically monitor anatomic variations and their dosimetric consequences during intensity modulated radiation therapy (IMRT) for head and neck (H&N) cancer by using a graphics processing unit (GPU)-based deformable image registration (DIR) framework. METHODS AND MATERIALS: Eleven IMRT H&N patients undergoing IMRT with daily megavoltage computed tomography (CT) and weekly kilovoltage CT (kVCT) scans were included in this analysis. Pretreatment kVCTs were automatically registered with their corresponding planning CTs through a GPU-based DIR framework. The deformation of each contoured structure in the H&N region was computed to account for nonrigid change in the patient setup. The Jacobian determinant of the planning target volumes and the surrounding critical structures were used to quantify anatomical volume changes. The actual delivered dose was calculated accounting for the organ deformation. The dose distribution uncertainties due to registration errors were estimated using a landmark-based gamma evaluation. RESULTS: Dramatic interfractional anatomic changes were observed. During the treatment course of 6 to 7 weeks, the parotid gland volumes changed up to 34.7%, and the center-of-mass displacement of the 2 parotid glands varied in the range of 0.9 to 8.8 mm. For the primary treatment volume, the cumulative minimum and mean and equivalent uniform doses assessed by the weekly kVCTs were lower than the planned doses by up to 14.9% (P=.14), 2% (P=.39), and 7.3% (P=.05), respectively. The cumulative mean doses were significantly higher than the planned dose for the left parotid (P=.03) and right parotid glands (P=.006). The computation including DIR and dose accumulation was ultrafast (∼45 seconds) with registration accuracy at the subvoxel level. CONCLUSIONS: A systematic analysis of anatomic variations in the H&N region and their dosimetric consequences is critical in improving treatment efficacy. Nearly real-time assessment of anatomic and dosimetric variations is feasible using the GPU-based DIR framework. Clinical implementation of this technology may enable timely plan adaptation and improved outcome.

A virtual phantom library for the quantification of deformable image registration uncertainties in patients with cancers of the head and neck.

PURPOSE: Deformable image registration (DIR) is being used increasingly in various clinical applications. However, the underlying uncertainties of DIR are not well-understood and a comprehensive methodology has not been developed for assessing a range of interfraction anatomic changes during head and neck cancer radiotherapy. This study describes the development of a library of clinically relevant virtual phantoms for the purpose of aiding clinicians in the QA of DIR software. These phantoms will also be available to the community for the independent study and comparison of other DIR algorithms and processes. METHODS: Each phantom was derived from a pair of kVCT volumetric image sets. The first images were acquired of head and neck cancer patients prior to the start-of-treatment and the second were acquired near the end-of-treatment. A research algorithm was used to autosegment and deform the start-of-treatment (SOT) images according to a biomechanical model. This algorithm allowed the user to adjust the head position, mandible position, and weight loss in the neck region of the SOT images to resemble the end-of-treatment (EOT) images. A human-guided thin-plate splines algorithm was then used to iteratively apply further deformations to the images with the objective of matching the EOT anatomy as closely as possible. The deformations from each algorithm were combined into a single deformation vector field (DVF) and a simulated end-of-treatment (SEOT) image dataset was generated from that DVF. Artificial noise was added to the SEOT images and these images, along with the original SOT images, created a virtual phantom where the underlying "ground-truth" DVF is known. Images from ten patients were deformed in this fashion to create ten clinically relevant virtual phantoms. The virtual phantoms were evaluated to identify unrealistic DVFs using the normalized cross correlation (NCC) and the determinant of the Jacobian matrix. A commercial deformation algorithm was applied to the virtual phantoms to show how they may be used to generate estimates of DIR uncertainty. RESULTS: The NCC showed that the simulated phantom images had greater similarity to the actual EOT images than the images from which they were derived, supporting the clinical relevance of the synthetic deformation maps. Calculation of the Jacobian of the "ground-truth" DVFs resulted in only positive values. As an example, mean error statistics are presented for all phantoms for the brainstem, cord, mandible, left parotid, and right parotid. CONCLUSIONS: It is essential that DIR algorithms be evaluated using a range of possible clinical scenarios for each treatment site. This work introduces a library of virtual phantoms intended to resemble real cases for interfraction head and neck DIR that may be used to estimate and compare the uncertainty of any DIR algorithm.

Radiosurgery technology development and use.

Radiosurgery first became a clinical option in the 1960's because of the Gamma Knife, and the technology proliferated in the 1980's due to the availability of linear accelerator radiosurgery. The technology has continued to develop with both Gamma Knife and linac radiosurgery due primarily to advances in computer technology and robotic automation. Many of these advances include planning systems that enhance the conformity of the dose distribution, and delivery systems that can more safely and efficiently delivery these more complex treatment plans. This manuscript details the evolution of technologies in stereotactic localization and delivery for intracranial radiosurgery.